Okay, but is it really restarting it if the old one remains? It's more like just making a new one. The main problem with this idea, I think, is to make people abandon the old one and switch to the new one. How to avoid stuff like another Bitcoin blahblahblah version appearing? And if we do it regularly, then some people will use blockchain #1 (probably the majority), others #2, #3 etc. This will split the community into small parts, and it'll just be confusing. Oh, and what about the prices? Wouldn't we technically just have different cryptocurrencies and thus each of them will have its own marketcap and price? I think it can have significant negative impact
What you describe is an engineering problem
You will likely find a lot of other technical issues if you start digging further and deeper. But this topic is not about such things, I'm more concerned with the idea itself taken on its own merits (and probably shortcomings). All in all, I think it is a good idea to keep the blockchain size to a minimum, and one way to actually get there is to reset or restart the blockchain regularly, say, every 2-3 years (terminology is a matter of convention here)